In recent years, the spreading of portable terminals, such as smart phones, has made it possible to use services on the network, such as an information search service and a ticket reservation service at any time. In addition, with an improvement in the performance of voice recognition technology and voice synthesis technology, electronic apparatuses, such as PCs (Personal Computers) or portable terminals, with a voice input function or a voice output function have been actively developed. For example, an input and output device, such as a headset capable of performing both an input operation using voice recognition and an output operation using voice synthesis, can be used to issue a request for processing to an electronic apparatus, such as a portable terminal or a PC, or to listen to the result of the processing even at the position separated from the electronic apparatus. Therefore, the user can obtain necessary information even when the user has no hands free or even when the user performs another operation.
However, a situation in which the user can perform an operation at the position away from the electronic apparatus which performs the actual processing is convenient for the user, but it is difficult to know the time when the message is notified by the system (electronic apparatus). Therefore, for example, during conversation, a response, such as a notice of the next meeting or the confirmation of the schedule, to the request from the user may return. When a voice during conversation (the voice of the user or the voice of the conversational partner) overlaps the message from the system (“system response”), the conversation is interrupted and it is difficult for the user to listen to the system response. Therefore, it is necessary to control the output time of the system response such that the conversation does not overlap the system response.
For example, JP-A 2009-20672 (KOKAI) discloses a technique in which a server apparatus which checks the work state of each user in real time presents the real-time work state of each user such that each user can check the state of a communication partner. However, this technique presents only the real-time work state of each user, but does not control the time when a message is output. In addition, for example, JP-A 2010-152119 (KOKAI) discloses a technique which performs control such that the speech of the user does not overlap the system response when the user communicates with the system. In this technique, when the user makes a speech, the system classifies the intention of the speech of the user and selects response candidates corresponding to the classified intention of the speech. A waiting time until a message corresponding to an answer sentence is output is set to each response candidate and the message corresponding to the answer sentence is output only when the user makes a new speech for the waiting time. However, in this technique, the message corresponding to the answer sentence is output only when the waiting time has elapsed, regardless of the current state of the user. Therefore, in some cases, the output time of the message is not appropriate.
In the above-mentioned techniques, it is difficult to notify the user of the message at an appropriate time.